No more Cloud Run, IAM, or egress fees.

≈ $95 USD/month

• vCPU (always-on)$62.21
• RAM (512 MB)$3.24
• Requesty (5 M)$2.0
• Egress (~10 GiB)$0.73
• Cloud Logging$0.0
• Static IP$1.46
• Ops/maintenance overhead$25.0

More realistic tends to be +20–40% because min-instances=1 leaves the CPU idle 95% of the day, but it's still billed.

≈ $73 USD/month · ISDOC invoicing

• vCPU + RAMincluded
• Requesty (5 M)included
• Egressnot charged
• Logy + auditincluded
• SSL + domainincluded
• Updates + monitoringincluded
• Ops time0 minutes/month

A fixed amount. No egress shock, no regional billing puzzle.

Bare-metal in the EU, not a hyperscaler markup

A Hetzner Tier III data center in Germany, dedicated CPU. GCP charges ~3× for the same performance because it bakes managed Kubernetes, multi-region replication, and BigQuery integration into the price — features you don't need for an sGTM proxy.

Shared infrastructure across tenants

One Hetzner box serves 200+ tenants (the sGTM container runs in Docker with vCPU caps + memory limits). On Cloud Run you pay for idle CPU 24/7 — min-instances=1 is the default. We multiplex it.

No egress, no logging fees

Hetzner includes 20 TB of outbound traffic. GCP's Cloud Logging charges $0.50/GB beyond 50 GB — for a busy store during Black Friday that's +CZK 500. We write logs to disk + 90-day retention, free.

You gain

• Fixed CZK invoicing — no monthly puzzle of "how much did it cost us this year".
• EU-only data residency — an EU Tier III data center (Germany). No US-parent puzzle for corporate clients.
• Free egress — no traffic-spike billing shock.
• Auto SSL + auto Container Config rollouts — no terraform/IAM lock-in.
• Sklik + Heureka native integrations — critical for the CZ market, on GCP you have to write them yourself.
• Czech support — a response during EU business hours, not through a Google Premium Support contract.
• ISDOC invoicing — standard for accountants in CZ/SK, GCP invoices in USD without ISDOC.

What you lose (honestly)

• Multi-region failover — DataNostro runs in one EU location; GCP Cloud Run can do multi-region. A non-issue for 99.9% of CZ/EU stores, but worth mentioning.
• Custom autoscaling rules — we scale automatically, you can't set "keep 10 instances on Black Friday". If you need enterprise scale with your own rules, we have a dedicated tier.
• BigQuery integration inside the Cloud Console — GCP can do a direct BigQuery export with one click. With us, export is via the API or a CSV dump from the dashboard.
• VPC peering / Cloud Armor / IAP — Google Cloud enterprise security features, which we replace with a cloud-level firewall + nginx rate-limiting.

If you need any of this, write to us — we'll help you decide honestly, even if it means: stay on GCP, it makes sense.

AI Debug Agent

Click „Explain the failures“ → Gemini goes through the last 50 failed events + your tenant config + Power-Ups → returns a likely cause + suggested fix. A confidence chip. No manual grepping through logs.

On GCP: write it yourself or a BigQuery query.

Data Quality Score

A daily 0-100 score with an A/B/C/D/F grade + a 30-day sparkline trend. Weights: match rate 30%, error freedom 30%, cookie 15%, latency 15%, consent 10%.

On GCP: Looker Studio from scratch.

Anomaly Detection + AI

A rolling 7d baseline detects „match rate dropped 18% vs the average“. AI immediately explains the cause („Meta token expired“). Pushed to Slack/Discord in parallel.

On GCP: Cloud Monitoring + Pub/Sub + Vertex AI.

Slack / Discord / Teams

Per-platform format (Block Kit / embed / MessageCard). Severity + category filters. A test button. Marketing teams live in Slack, IT in Discord — not in an email inbox.

On GCP: a Cloud Function with your own code.